Categories Might Be Safe, But Not Vendors
A follow-up discussion of AI, SaaS, and EdTech survival
Phil Hill
April 07, 2026
Was this forwarded to you by a friend? Sign up, and get your own copy of the news that matters sent to your inbox every week. Sign up for the On EdTech newsletter. Interested in additional analysis? Upgrade to the On EdTech+ newsletter.
The response to last week's piece on AI, SaaS, and EdTech Survival generated more direct messages and LinkedIn commentary than almost anything I've published this year, which indicates the question is live in a way that the broader public debate hasn't quite caught up to yet. Most of that debate is still stuck on the wrong frame: what does AI replace? The more useful question, and the one several sharp readers explored further, is what AI specifically cannot replace—and why that distinction matters more than the survival question itself.
Let’s broaden the discussion beyond my response to the HackerNoon article that focused on open source, and look further at the implications of this can’t replace question for AI in EdTech.
Product Definition vs. Coding
Marcus Popetz captured one key issue quite well in a LinkedIn comment: knowing what software to build — effective product definition — has not been commoditized by AI.
The hard part isn't building it, it's knowing what to build. That's why "product market fit" is the goal that companies chase. And you might mistakenly assume that people in a role know what they need to the level to specify it but they don't.
